UDP-N-acetylglucosamine:.beta.-D-mannoside .beta.l -4N-acetyl-glucosaminyltransferase III (EC 2.4.1.144: hereinafter referred to simply as GnT-III), which is an enzyme that transfers a GlcNAc residue in UDP-N-acetylglucosamine (UDP-GlcNAc) to a mannose (Man) residue forming a .beta.l-4 bond in an asparagine binding type sugar chain, was reported for the first time by Narasimhan see Journal of Biological chemistry, 257, 10235-10242 (1982)!. The GlcNAc transferred by GnT-III, which is called a bisecting GlcNAc, has been found in the sugar chains of various glycoproteins. It has been further reported by Narasimhan et al. see J. Biol. Chem., 263, 1273-1282 (1988)! and Pascale et al. see Carcinogenesis, 10, 961-964 (1989)! that the activity of GnT-III increases in rat liver accompanying its canceration.
In addition, Ishibashi et al. have reported an increase in the activity of GnT-III in the serum of a human patient with hepatic cancer see Clinica Chimica Acta, 185, 325-332 (1989)!.
Regarding genes, furthermore, a gene coding for GnT-III originating in rat (rat GnT-III) has been isolated by one of the present inventors see Japanese Patent Application No. 69345/1992!. It is believed that metastasis is the main problem in clinical medicine for cancer. Namely, metastasis means a phenomenon where cancer cells from a primary tumor enter into the blood system or the lymphatic system and form a new tumor in another part of the body via such a system. If cancer cells do not metastasize or the cancerous metastasis can be prevented, a patient with cancer can be saved by excising the cancer.
It has been clarified that, in many solid cancers, sugar chain structures expressed on the surfaces of cancer cells vary as the cancer advances, i.e., depending on the stage of the advance and the occurrence of the metastatic character. It is therefore considered that the sugar chain structures on the surface of the cancer cells vary as the cancer advances and is acquire metastatic character. In particular, it has been widely known that a cell having a sugar chain with a specific branched structure expressed thereon has a potent ability to metastasize as reported by Dennis et al. Science, 236, 582-585 (1987)!. The evidence thereof is as follows:
1) A leukoagglutinin originating in kidney bean (L-PHA) recognizes an asparagine linked sugar chain having a Gal.beta.l-4GlcNAc.beta.l-6(Gal.beta.l-4GlcNAc.beta.l-2)- Man.alpha.l branched structure and binds thereto. A mouse cancer cell line MDAY-D2, which shows a sensitivity to this L-PHA, has metastatic potential. On the other hand, a cell line showing a resistance to L-PHA has low metastatic potential. 2) An L-PHA binding type glycoprotein is detected from the cell membrane of a metastatic cell line, while no L-PHA binding type glycoprotein is detected from the cell membrane of a cell line having low metastatic potential. The incidence of the metastasis correlates to the occurrence of the L-PHA binding type glycoprotein.
3) When an oncogene is introduced into rat cells, the L-PHA binding type glycoprotein appears therein. When these cells are injected into a nude mouse, a tumor is formed and metastasizes.
The results of tissue staining with L-PHA indicate that the L-PHA binding type sugar chain appears not only in cancer cells which have been experimentally formed but also in human breast cancer and colon cancer and that the intensity of the L-PHA staining is elevated as the cancer advances Cancer Research, 51, 718-723 (1991)!.
Since a sugar chain is not a direct product of a gene, a change in the sugar chain structure depends on a glycosyltransferase. It has been reported that a mouse cell line having the L-PHA binding type sugar chain expressed thereon and showing metastatic potential has a higher activity of N-acetyl-glucosaminyltransferase (GnT) V (hereinafter referred to simply as GnT-V), which forms GlcNAc.beta.l -6Man.alpha.l branching, than a cell line showing low metastatic potential Science, 236, 582.beta.585 (1987)!. Also, the intensity of L-PHA staining positively correlates to the GnT-V activity in a human breast cancer tissue Cancer Research, 49, 945-950 (1989)!.
As described above, there have been detailed studies on the relationship between the appearance of the L-PHA binding type sugar chain having the GlcNAc.beta.l-6Man.alpha.l-branched structure and the metastatic character of cancer cells. However it has not been revealed so far whether or not the sugar chain on the surface of the cell having such a structure specifies the extent of the advance of the cancer and directly causes the acquisition of the metastatic character. Needless to say, there has been developed neither a method nor a drug whereby the possibility of cancerous metastasis can be effectively reduced.
In the process of the studies relating to structural changes in sugar chains on the surface of cells, the present inventors have successfully acquired rat and human GnT-III genes (Japanese Patent Laid-Open No. 38767/1994 and U.S. patent appln. Ser. No. 08/107,173). This enzyme forms the GlcNAc.beta.l-4Man.beta.l structure of an asparagine linked sugar chain, i.e., the so-called bisecting GlcNAc. It has been reported that the activity of this GnT-III also increases in cancer cells. In particular, the present inventors have clarified that the activity of this enzyme increases in rat or human liver cancer tissues or the serum of a patient with liver cancer Biochemical and Biophysical Research Communications, 152, 107-112 (1988); and Clinica Chimica Acta, 185, 325-332 (1989)!. Regarding cancers other than liver cancer, it has also been reported that GnT-V and GnT-III activities increase in cells which have been malignantly transformed by introducing N-ras protooncogene thereinto Journal of Biological Chemistry, 266, 21674-21680 (1991)! and that GnT-III activity largely increases in metastatic prostatic cancer cells FEBS Letters, 308, 46-49 (1992)!.